Insights into hydrogen generation from formic acid using ruthenium complexes

Article abstract of DOI:10.1021/om900099u

The decomposition of a HCO₂H/Et₃N azeotrope to a mixture of hydrogen and carbon dioxide may be catalyzed by a number of Ru(III) and Ru(II) complexes with high efficiency at ca. 120 °C. Evidence that suggests that the precatalyst may

Full text of DOI:10.1021/om900099u

Organometallics 2009, 28, 4133–4140 4133
DOI: 10.1021/om900099u
Insights into Hydrogen Generation from Formic Acid Using Ruthenium
Complexes
David J. Morris, Guy J. Clarkson, and Martin Wills*
Department of Chemistry, The University of Warwick, Coventry, CV4 7AL U.K.
Received February 9, 2009
The decomposition of a HCO2H/Et3N azeotrope to a mixture of hydrogen and carbon dioxide may
be catalyzed by a number of Ru(III) and Ru(II) complexes with high efficiency at ca. 120 °C. Evidence
that suggests that the precatalyst may in each case be a common ruthenium dimer has been obtained
through 1H NMR and X-ray crystallographic studies of the complexes formed in situ and of analysis
of the gases generated in the reaction using FTIR and gas chromatography methods.
There is currently a high level of interest in the use of
hydrogen as an energy vector; since the only byproduct of its
combustion is water, it represents a clean power source.1,2
Currently, hydrogen is generated from fossil fuel sources
using steam reforming and related processes.3 Heteroge-
neous catalytic systems have been investigated for hydrogen
generation from biomass, and this continues to be an im-
portant area for future research work.4
Formic acid (FA) may be regarded as a source of, and
storage material for, hydrogen.5-12 FA will decompose to H2/
CO2 in a thermodynamically favored process (Scheme 1),5k
although highly elevated temperatures are required.5a-c The
process may be catalyzed by a number of metal complexes.
Publications from the 1950s through to the present day5d-q
describe the use of various metals including Au/Pd,5d,5e Ag,5f
Ni,5f,5g,5i Cu,5g,5i Pt,5h,5p,5q Ru,5h,5o Ir,5h Ti, V, Cr, Mn, Fe,
Co,5h W,5j Rh,5k,5n Pt,5l and Cd.5m Formate salts may also be
used as effective sources of hydrogen, including ammonium
and metal formates.6 The reverse reaction, i.e., the hydro-
genation of carbon dioxide, has been demonstrated to be a
practical and viable method of making FA.7 Fuel cells based
on FA have been developed and are being commercialized.8
Theintermediacy of FAinthe water gas shiftreactionpresents
the possibility of making hydrogen from water and carbon
monoxide, via FA.9 Hence, there is significant potential for
the development of a viable “joined up” system of hydrogen
*Corresponding author. Fax: (þ44) 24 7652 3260. Tel: (þ44) 24 7652
4112. E-mail: m.wills@warwick.ac.uk.
(1) (a) Hydrogen Energy, Challenges and Prospects; Rand, D. A. J.,
Dell, R. M., Eds.; RSC Publishing: Cambridge, 2007. (b) Hydrogen as a
Future Energy Carrier; Z€uttel, A., Borgschulte, A., Schlapbach, L., Eds.;
Wiley VCH: Weinheim, 2008. (c) Bockris, J. O. M. Science 1972, 176, 1323.
(d) Crabtree, G. W.; Dresselhaus, M. S.; Buchman, M. V. Phys. Today 2004,
57, 39–44. (e) Baykara, S. Z.; Int. J. Hydrogen Energy 2005, 30, 545–553.
(f) Z€uttel, A.; Schlapbach, L. Nature 2001, 414, 353–358.
ꢀ
(2) van den Berg, A. W. C.; Arean, C. O. Chem. Commun. 2008, 668–
681.
~
(3) Navarro, R. M.; Pena, M. A.; Fierro, J. L. G. Chem. Rev. 2007,
107, 3952–3991.
(4) (a) Dupont, V.; Ross, A. B.; Hanley, I.; Twigg, M. V. Int. J.
Hydrogen Energy 2007, 32, 67–79. (b) Ramírez de la Piscina, P.; Homs, N.
Chem. Soc. Rev. 2008, 37, 2459–2467. (c) Zhou, C.-H.; Beltramini, J. N.;
Fan, Y.-X.; Lu, G. Q. (M.) Chem. Soc. Rev. 2008, 37, 527–549.
(d) Einashaie, S.; Chen, Z.; Prasad, P. Int. J. Green Energy 2007, 4,
249–282.
(7) (a) Jessop, P. G.; Ikariya, T.; Noyori, R. Chem. Rev. 1995, 95, 259–
272. (b) Leitner, W. Angew. Chem., Int. Ed. 1995, 34, 2207–2221.
(c) Jessop, P. G.; Hsiao, Y.; Ikariya, T.; Noyori, R. Chem. Commun. 1995,
707–708. (d) Leitner, W.; Dinjus, E.; Gabsser, F. J. Organomet. Chem. 1994,
475, 257–266. (e) Jessop, P. G.; Ikariya, T.; Noyori, R. Nature 1994, 368,
(5) (a) Hinshelwood, C. N.; Hartley, H. J. Chem. Soc. Trans. 1923,
123, 1333–1338. (b) Wakai, C.; Yoshida, K.; Tsujino, Y.; Matubayashi, N.;
Nakamura, M. Chem. Lett. 2004, 33, 572–573. (c) Blake, P. G.; Davies,
H. H.; Jackson, G. E. J. Chem. Soc. (B) 1971, 1923–1925. (d) Eley, D. D.
Leutic, P. Trans. Faraday Soc. 1957, 53, 1483–1487. (e) Zhou, X.; Huang, Y.;
Xing, W.; Liu, C.; Liao, J.; T. Lu, T. Chem. Commun. 2008, 3540–3542.
(f ) Tamaru, K. Trans. Faraday Soc. 1959, 55, 824–832. (g) Duell, M. J.
Robertson, A. J. B. Trans. Faraday Soc. 1961, 57, 1416–1423. (h) Coffey, R. S.
Chem. Commun. 1967, 923–924. (i) Inglis, H. S.; Taylor, D. J. Chem. Soc.
(A) 1969, 2985–2987. ( j) Bhattacharya, A. K. J. Chem. Soc., Faraday
Trans. 1 1979, 75, 863–871. (k) Strauss, S. H.; Whitmire, K. H.; Shriver, D. F.
J. Organomet. Chem. 1979, 174, C59–C62. (l) Paonessa, R. S.; Trogler, W. C. J.
Am. Chem. Soc. 1982, 104, 3529–3530. (m) Willner, I.; Goren, Z. Chem.
Commun. 1986, 172–173. (n) King, R. B.; Bhattacharyya, N. K. Inorg.
Chim. Acta 1995, 237, 65–69. (o) Gao, Y.; Kuncheria, J.; Yap, G. P. A.;
Puddephatt, R. J. Chem. Commun. 1998, 2365–2366. (p) Hyde, J. R.;
Poliakoff, M. Chem. Commun. 2004, 1482–1483. (q) Garcia-Verdugo, E.;
Liu, Z.; Ramirez, E.; Garcia-Serna, J.; Frago-Dubreuil, J.; Hyde, J. R.;
Hamley, P. A.; Poliakoff, M. Green Chem. 2006, 8, 359–364.
ꢀ
ꢀ
231–233. (f ) Elek, J.; Nadasdi, L.; Papp, G.; Laurenczy, G.; Joo, F. App.
Catal. A: Gen. 2003, 255, 59–67. (g) Fukuzumi, S. Eur. J. Inorg. Chem.
2008, 1351–1362.
(8) (a) Ha, S.; Dunbar, Z.; Masel, R. I. J. Power Sources 2006, 158,
129–136. (b) Ha, S.; Larsen, R.; R. I. Masel, R. I. J. Power Sources 2005,
144, 28–34. (c) Zhu, Y.; Ha, S.; Masel, R. I. J. Power Sources 2004, 130, 8–
14. (d) Ha, S.; Adams, B.; Masel, R. I. J. Power Sources 2004, 128, 119–24.
(9) (a) Yoshida, T.; Ueda, Y.; Otsuka, S. J. Am. Chem. Soc. 1978, 100,
3941–3942. (b) Yoshida, K.; Wakai, C.; Matubayashi, N.; Nakahara, M. J.
ꢀ
Phys. Chem. 2004, 108, 7479–7482. (c) Joo, F. ChemSusChem 2008, 1,
805–808. (d) Enthaler, S. ChemSusChem 2008, 1, 801–804.
(10) (a) Boddien, A.; Loges, B.; Junge, H.; Beller, M. Angew. Chem.,
Int. Ed. 2008, 47, 3962–3965. (b) Boddien, A.; Loges, B.; Junge, H.; Beller,
M. ChemSusChem 2008, 1, 751–758. (c) Junge, H.; Boddien, A.; Capitta, F.;
Loges, B.; Noyes, J. R.; Gladiali, S.; Beller, M. Tetrahedron Lett. 2009, 50,
1603–1606.
(11) (a) Fellay, C.; Dyson, P. J.; Laurenczy, G. Angew. Chem., Int. Ed.
2008, 47, 3966–3968. (b) Fellay, C.; Yan, N.; Dyson, P. J.; Laurenczy, G.
Chem.;Eur. J. 2009, 15, 3752–3760.
(6) (a) Delpuech, J. J.; Ducom, J.; Michon, V. Chem. Commun. 1970,
1187–1188. (b) Wiener, H.; Sasson, Y.; Blum, J. J. Mol. Catal. 1986, 35,
277–284. (c) Onsager, O.-T.; Brownrigg, M. S. A.; Lodeng, R. Int. J.
Hydrogen Energy 1996, 21, 883–885. (d) Fukuzumi, S.; Kobayashi, T.;
Suenobo, T. ChemSusChem 2008, 1, 827–834.
(12) Wagner, K. Angew. Chem., Int. Ed. 1970, 9, 50–56.
r
2009 American Chemical Society
Published on Web 07/02/2009
pubs.acs.org/Organometallics

4134 Organometallics, Vol. 28, No. 14, 2009
Morris et al.
Scheme 1
generation, storage, and transport based on the use of FA,
which is safer to use than high-pressure hydrogen gas.
Despite extensive prior research work, the rate at which
hydrogen may be generated by the FA decomposition reac-
tion still remains unoptimized. Recent papers, published
during the course of our own studies, have highlighted this
issue and presented potential solutions to it in the form of
highly active homogeneous Ru catalysts.10,11 Laurenczy has
reported a continuous-injection system in which a FA/
sodium formate solution is degraded to CO2 and H2 (without
CO contamination) using [Ru2(H2O)6(OTs)2] with a water-
soluble phosphorus ligand. The best results were achieved at
120 °C.11 Beller10 chose to investigate the formation of
hydrogen from FA-amine adducts (FA/triethylamine (FA/
TEA) is known to form a stable 5:2 azeotrope12). Using a
series of Ru(III) and Ru(II) catalysts, with tertiary amines,
the combination of [RuCl2(PPh3)3] and Et3N emerged as the
best. At 40 °C, a TOF (turnover frequency; moles hydrogen/
moles catalyst/h) of 3630 h-1 (first 20 min) was achieved and
represents the highest reported activity for FA decomposi-
tion at this temperature.10 While this is likely to remain
unchallenged for the 40 °C reaction, in this paper we describe
a catalyst system for hydrogen generation from FA with
TOFs of up to 1.80 ꢀ 104 h-1 at 120 °C.
Figure 1. Hydrogen and carbon dioxide generation using cata-
lyst 1 (6.3 mg in 2.5 mL 5:2 FA/TEA, 0.0088 mmol), at rt (20 °C).
Scheme 2. Proposed Mechanism of Hydrogen Generation from
FA by Catalyst 1
and turnover frequency (TOF) could thus be readily calcu-
lated. The results obtained from using catalyst 1 are shown
in Figure 1. The maximum TOF was ca. 490 h-1 after
ca. 35 min, and the volume of consumed FA at the end of
the reaction was 7.3 mmol (335 mg, 0.28 mL); hence only
ca. 20% had been consumed. This indicated that either
catalyst deactivation is taking place or the ratio of FA to
triethylamine must be above a minimal level for the catalyst
to be active. Although this clearly indicates a limitation, the
maximum TOF was very high for a catalytic decomposition
of FA at room temperature.
Control reactions were carried out in order to determine
whether the full catalyst structure was required for hydrogen
generation (see Supporting Information). It was also ob-
served that FA alone does not work; a basic additive is
required, in accord with Beller’s findings.10 Aqueous FA and
sodium formate were tested and gave no evolution of gas.
The analogous ruthenium(II) catalyst 2 was less active in this
application.
Our proposed mechanism of the hydrogen formation is
anticipated to be closely related to that of ATH of ketones13
(Scheme 2). In the first step, HCl is eliminated from 1 to
furnish the 16-electron species 3, which abstracts two hydro-
gen atoms from FA13 to form ruthenium hydride complex 4.
In the absence of substrate, to which hydrogen would
normally be transferred, competitive hydrogen elimination
operates instead. The rate of gas evolution decreases sharply
in the presence of a ketone substrate.
In view of the observed results, we speculated that other
rhodium and ruthenium complexes containing a heteroatom
adjacent to the metal may be able to operate through a
similar process and thus generate hydrogen from FA/TEA.
After a survey of a range of candidate organometallic com-
plexes, we found that the commercially available complex
[RuCl2(DMSO)4] (5) was capable of the production of
hydrogen from FA with a very high TOF (up to ca. 1.80 ꢀ
104 h-1) and TON (2.50ꢀ104 after 4 cycles of FA replenish-
ment) and without deactivation, at 120 °C. Other metal
complexes (TONs in parentheses), including Rh2(OAc)4
We have studied the process of asymmetric transfer hydro-
genation (ATH) with Ru and Rh catalysts,13-15 using FA/
TEA as the hydrogen source, a process first reported by
Noyori et al.13 During the course of our studies on closely
related but highly active tethered catalysts such as 1 and 2, we
frequently observed substantial gas generation (at rt) from
FA/TEA during the course of the reaction. This was also
noted to be hydrogen gas by Noyori in his 1995 paper14b and
subsequently by other researchers in this area.15 We also
have established that the gas produced by 1, 2, and other
catalysts we have studied is a mixture of H2 and CO2 (see
Experimental Section and Supporting Information).
Given the resurgence of interest in FA as a hydrogen-
carrying vector,5e,10,11 we elected to study this reaction
further. The rate of hydrogen generation from 5:2 (molar)
FA/triethylamine azeotrope (FA/TEA) was accurately de-
termined using a gas syringe to directly measure the volume
produced in the reaction (see Supporting Information). The
turnover number (TON: moles hydrogen/moles catalyst)
(13) (a) Hashiguchi, S.; Fujii, A.; Takehara, J.; Ikariya, T.; Noyori, R.
J. Am. Chem. Soc. 1995, 117, 7562–7563. (b) Fujii, A.; Hashiguchi, S.;
Uematsu, N.; Ikariya, T.; Noyori, R. J. Am. Chem. Soc. 1996, 118, 2521–
2522. (c) Koike, T.; Ikariya, T. Adv. Synth. Catal. 2004, 346, 37–41. (d)
Ikariya, T.; Murata, K.; Noyori, R. Org. Biomol. Chem. 2006, 4, 393–406. (e)
Gladiali, S.; Alberico, E. Chem. Soc. Rev. 2006, 36, 226–236.
(14) (a) Hayes, A. M.; Morris, D. J.; Clarkson, G. J.; Wills, M. J. Am.
Chem. Soc. 2005, 127, 7318–7319. (b) Hannedouche, J.; Clarkson, G. J.;
Wills, M. J. Am. Chem. Soc. 2004, 126, 986–987. (c) Matharu, D. S.; Morris,
D. J.; Kawamoto, A. M.; Clarkson, G. J.; Wills, M. Org. Lett. 2005, 7, 5489–
5491. (d) Matharu, D. S.; Clarkson, G. J.; Morris, D. J.; Wills, M. Chem.
Commun. 2006, 3232–3234.
(15) (a) Himeda, Y.; Onozawa-Komatsuzaki, N.; Sugihara, H.;
Arakawa, H.; Kasuga, K. J. Mol. Catal. A: Chem. 2003, 195, 95–100.
(b) Soleimannejad, J.; Sisson, A.; White, C. Inorg. Chim. Acta 2003, 121–
128.

Article
Organometallics, Vol. 28, No. 14, 2009 4135
continue to produce hydrogen over several cycles of added
formic acid, without deactivation. The use of a higher
concentration of catalyst results in faster generation of gas,
but the TOFs were essentially unchanged (see Supporting
Information), and the hydrogen gas from the reaction could
be used directly in a hydrogen PEM fuel cell (see Supporting
Information).
The observation of similar and increasing activity over the
first three cycles indicates that a common catalyst is forming
at high temperatures from each starting complex. There is
evidence in the literature that dinuclear, divalent (Ru(II)Ru-
(III)) complexes containing bridging formate ligands are
formed under similar conditions to ours.5o,16 The com-
plexes [Ru2(HCO2)4Cl 2H2O], [Ru2(HCO2)4Cl2]-, and [Ru2-
3
(HCO2)4Cl.2H2O] BF4 each contain four bridging formate
3
ligands between bonded ruthenium atoms, which are
formally a combination of Ru(II) and Ru(III).16a-c The
complex [Ru2(HCO2)4Cl2]- has been characterized by X-ray
crystallography,16c and the electronic structure of several such
complexes has been described in detail.16d,16f,16g Closely related
complexes containing bridging acetate ligands are also known
and have been characterized.16c,17 A diphosphine-bridged dir-
uthenium complex, [Ru2(μ-CO(CO)4(μ-dppm)2], has been re-
ported to be active in formic acid decomposition to hydrogen
and carbon dioxide.5o,18 We speculated that a formate-bridged
dimer may be formed in this process and subsequently parti-
cipate in the dehydrogenation process via a formate addition
intermediate. This would account for the requirement for TEA
in the reaction, and, by implication, an ammonium formate salt.
In order to gain evidence to support the proposed forma-
tion of a Ru dimer, we attempted to isolate an active catalytic
material from the reaction; however this proved unsuccess-
ful. The addition of substoichiometric quantities of triphe-
nylphosphine, however, provided some insights (Figure 3).
The addition of 0.25 equiv of PPh3 to the reaction after one
cycle of formic acid consumption gave a slight increase in
activity in the second cycle, analogous to the normal pattern,
while addition of a second 0.25 equiv had little effect. The
addition of a third 0.25 equiv, however, resulted in degrada-
tion of the rate of hydrogen production, while addition of a
full equivalent of PPh3 resulted in inhibition of the reaction.
These results are somewhat in contrast to those obtained
by Beller,10 in which the addition of a phosphine is shown to
increase the rate of FA decomposition (in the presence of an
amine); however it should be noted that Beller’s results were
obtained at a lower temperature than ours and may reflect
the formation of a different catalytic species. From the reaction
Figure 2. Hydrogen and carbon dioxide generation using Ru-
(II) and Ru(III) catalysts (TEA/FA, 120 °C, oil bath; see
Supporting Information for full data). Maximum TOFs are
1.78 ꢀ 104 [Ru2Cl2(DMSO)4], 1.80 ꢀ 104 [RuCl2(NH3)6], 1.74 ꢀ
104 [RuCl3], 1.78 ꢀ 104 [Ru2(HCO2) 2(CO)4].
(17), RhCl3 (17), IrCl3 (434), PdCl2 (4), CoCl2 (3), FeCl3 (3),
FeCl2 (3), NiCl2 (3), and (EN)2PtCl2 (14), were much less
effective, although (NH3)2OsCl6 (1.30ꢀ 103) showed some
promise (see Supporting Information). The results of a
typical experiment with complex 5 are shown in Figure 2,
which illustrates the instantaneous TOF for the reaction.
Illustrating the progress of the reaction in this way gives an
immediate picture of the efficiency of the catalyst at any
moment in the catalytic cycle. After an initial induction
period10 the TOF increased rapidly, before decreasing after
a maximum of ca. 1.50 ꢀ104 h-1 as FA was consumed.
Several further aliquots of FA (calculated to replace the
volume lost by gas formation) were added at the minimum-
TOF point in the cycle, and in each case the TOF pattern was
repeated, in each cycle rising to a higher maximum level. These
results suggest that the catalyst is undergoing further activa-
tion in each cycle. The average TOF over this reaction period is
ca. (1.0-1.2)ꢀ104 h-1. If a continuous feed system were to be
established, i.e., to keep the TOF at the highest point, then the
average TOF has the potential to exceed 1.60 ꢀ 104 h-1
.
Over the course of the above reaction, each portion of
added FA released ca. 3.5-4 L of total gas, i.e., 1.75-2 L of
hydrogen. In the runs in Figure 2, each 3.5-4 L gas cycle was
completed in ca. 20 min. Careful checks were conducted
using cooled tubing to ensure that temperature changes were
not giving false positive readings. It was found that the
addition of extra DMSO (20 equiv) had no effect on the
reaction; however the use of TEA was essential.10 Other
amines, for example, dimethyloctylamine, may be used, but
provide no advantage over TEA.10 An NMR investigation
was conducted by sampling the liquid from the reaction over
the course of one reaction cycle (see Supporting Informa-
tion). The graph of these results revealed that the ratio of FA
to TEA remained above the theoretical value for that part of
the cycle, with an inflection that indicated that the FA:TEA
ratio for the majority of the hydrogen-generation cycle was
ca. 2:1. This suggests that some of the TEA may be prefer-
entially evaporating at such a rate that the remaining liquid
maintains a fairly constant FA:TEA ratio.
(16) (a) Mukaida, M.; Nomura, T.; Ishimori, T. Bull. Chem. Soc. Jpn.
1967, 45, 2143–2147. (b) Mukaida, M.; Nomura, T.; Ishimori, T. Bull. Chem.
Soc. Jpn. 1967, 40, 2462. (c) Bino, A.; Cotton, F. A.; Felthouse, T. R. Inorg.
Chem. 1979, 18, 2599–2604. (d) Norman, J. G.; Renzoni, G. E.; Case, D. A.
J. Am. Chem. Soc. 1979, 101, 5256–5267. (e) Lehmann, H.; Wilkinson, G.
J. Chem. Soc., Dalton Trans. 1981, 191–195. (f ) Clark, R. J. H.; Ferris, L. T. H.
Inorg. Chem. 1981, 20, 2759–2766. (g) Miskowski, V. M.; Loehr, T. M.;
Gray, H. B. Inorg. Chem. 1988, 27, 4708–4712.
(17) (a) Stephenson, T. A.; Wilkinson, G. J. Inorg. Nucl. Chem. 1966,
28, 2285–2291. (b) Mitchell, R. W.; Spencer, A.; Wilkinson, G. J. Chem.
Soc., Dalton Trans. 1973, 846–854. (c) Bianchi, M.; Frediani, P.; Matteoli,
U.; Menchi, G.; Piacenti, F.; Petrucci, G. J. Organomet. Chem. 1983, 259,
207–214. (d) Auzias, M.; Therrien, B.; Suess-Fink, G. Inorg. Chim. Acta.
2006, 356, 3412–3416.
(18) (a) Gao, Y.; Kuncheria, J. K.; Jenkins, H. A.; Puddephatt, R. J.;
Yap, G. P. A. J. Chem. Soc., Dalton Trans. 2000, 3212–3217. (b) Almeida
Further investigation also revealed that anhydrous RuCl3
(max. TOF 1.74 ꢀ 104) and [RuCl2(NH3)6] (max. TOF
1.80 ꢀ 104) performed as well as the DMSO complex in this
application (also illustrated in Figure 2). Both catalysts
~
Lenero, K.; Kranenburg, M.; Guari, Y.; Kamer, P. C. J.; van Leeuwen,
P.W. N. M.; Sabo-Etienne, S.; Chaudret, B. Inorg. Chem. 2003, 42, 2859–
2866. (c) Gao, Y.; Jennings, M. C.; Puddephatt, R. J. Organometallics 2001,
20, 1882–1888.

4136 Organometallics, Vol. 28, No. 14, 2009
Morris et al.
Figure 3. Observed TOFs using [RuCl2(DMSO)4] with PPh3 added in 0.25 equiv portions after (i) 40, 80, and 150 min (10 mL FA/TEA,
[RuCl2(DMSO)4] (6.5 mg, 13.4 μmol), 120 °C, oil bath).
Figure 4. X-ray crystallographic structure of [Ru2(HCO2)2(CO)4(PPh3)2], 6.
in which PPh3 was added to [RuCl2(DMSO)4] in FA/TEA, a
distinctive peak was observed in the 31P NMR spectrum at
δ12.6 ppm, suggesting the formation of a new complex.
Triphenylphosphine oxide formation was also observed at
δ 28.8 ppm in the 31P NMR spectrum. A much cleaner
formation of the new complex, with minimal accompanying
Ph3PO, was however achieved using ruthenium trichloride
hydrate under the same conditions, and it was possible to
isolate the new complex from the reaction as yellow crystals.
X-ray crystallography revealed the structure of the novel
complex [Ru2(HCO2)2(CO)4(PPh3)2], 6 (Figure 4); in this com-
plex, a dimer of ruthenium with two bridging formates and a
phosphine located on each metal has formed. It is also clear that
decomposition of formate to carbon monoxide has taken place
during the activation process in which the catalyst is formed.
A related acetate-bridged system has been reported, which was
formed from the reaction of [Ru2(CH3CO2)4(CO)4] with trialk-
ylphosphines;17c however we are not aware of a synthesis of
such a complex directly from ruthenium trichloride. Complex 6
was also prepared in larger quantities, thereby permitting full
characterization, via the formation of known complex [Ru2-
(HCO2)2(CO)4] (from dodecacarbonyltriruthenium)19 followed
by addition of triphenylphosphine in THF (reflux, 6 h; Sup-
porting Information). The use of pure complex 6 for hydrogen
generation from FA/TEA under the conditions described
above (120 °C, oil bath) proved to be successful, although the
TOF observed was lower (ca. 5.2ꢀ103 h-1) than those seen for
the phosphine-free reaction.
The isolation of an intermediate organometallic species
that reflected the situation when<1 equiv of PPh3 was added
proved to be rather more challenging. However, by adding
just under 1 equiv (relative to Ru metal) of triphenylpho-
sphine to [Ru2(HCO2)4(CO)4]19 at reflux in THF for 1 h, it
was possible to form the novel complex [Ru4(HCO2)4-
(CO)8(PPh3)2], 7 (Figure 5), in which a Ru tetramer is
coordinated by phosphines at the ends and contains bridging
formate ligands. As was the case for 6, two formates form a
bridge between pairs of ruthenium atoms, and the structure
contains a symmetrical “head to tail” bridging structure in
which a “(RuO)2” “square” is formed. Thus complex 7 is
essentially a dimer of the complex that would be formed by
loss of phosphine from 6 and contains a 2:1 ratio of ruthe-
nium to phosphine. Complex 7 proved to be unstable and
could not be characterized by mass spectrometry due to
rapid disproportionation to 6 and (presumed) [Ru2(HCO2)2-
(CO)4] during ionization. This disproportionation also
appears to take place in solution (followed by 31P NMR),
(19) Crooks, G. R.; Johnson, B. F. G.; Lewis, J.; Williams, I. G.;
Gamien, G. J. Chem. Soc. (A) 1969, 2761–2769.

Article
Organometallics, Vol. 28, No. 14, 2009 4137
Figure 5. X-ray crystallographic structure of [Ru4(HCO2)4(CO)8(PPh3)2], 7.
and as a result, complex 7 could not be used in hydrogen
generation experiments with confidence. Hence it is not
possible to confirm whether 7 is an active species in the
formic acid decomposition or merely an intermediate in the
formation of the relatively unreactive diphosphine complex
6. However, the complex is to our knowledge representative
of a new class of ruthenium tetramer structure.
The dimeric complex [Ru2(HCO2)4(CO)4] was also tested
Figure 6. Ruthenium complexes formed in these studies.
directly in the formic acid decomposition reaction (Figure 2)
and gave a maximum TOF of ca. 1.78 ꢀ 104 h-1. Both the
maximum TOF and the pattern of TOF versus gas volume
matched closely to those observed in the reactions using each
of the other complexes. Although [Ru2(HCO2)4(CO)4] could
not be isolated directly from the reactions of [Ru2Cl2(DM-
SO)4], [RuCl2(NH3)6], or [RuCl3], the more stable complex 6
was isolated upon addition of triphenylphosphine. This is
probably formed via trapping of [Ru2(HCO2)4(CO)4] during
the course of the reaction. Given the similarity of the profile
of gas generation using all four ruthenium sources, we
speculate that, under the conditions tested, the generation
of hydrogen may be taking place through formation of a
highly active common catalyst such as 8 at elevated tempera-
ture. A plausible mechanism for hydrogen generation would
involve the addition of formate (i.e., from triethylammo-
nium formate) to 8, followed by generation of a molecule of
hydrogen by the subsequent reaction with triethylammo-
nium cation and reclosure of the formate bridge (Scheme 3).
A Ru-H species, several of which are known to release
hydrogen gas, may be involved at some stage in the cycle.11,18
Addition of PPh3 results in formation of a quantity of 6,
which is much less catalytically active due to the blocking of
the ruthenium atoms by the large phosphine groups, but still
retains some low reactivity, possibly through a mechanism in
which the phosphine is dissociated (i.e., to give 9).
The formation of the active catalyst requires heating of
the reaction to over 100 °C possibly in order to promote the
formation of a CO-containing complex. At this tempera-
ture there is a distinctive color change in the solution from a
dull yellow to bright yellow, and rapid FA decomposition
is observed. Throughout this project, and particularly in
view of the formation of CO-containing complexes, efforts
were made to establish the composition of the gas (full
details are contained in the Supporting Information). IR
spectroscopy was used to analyze the gas by taking samples
at various times in the FA decomposition cycle, and this
technique permitted the clear identification of carbon
dioxide and carbon monoxide. Using an FT-IR method,
Scheme 3. Proposed Catalytic Cycle in H2 Generation from FA
Using Ruthenium Complexes
we were able to establish that the FA decomposition gases
from the [RuCl2(DMSO)4]-precatalyst reaction contained
ca. 0.02 mol % CO (200 ppm). A more accurate measure-
ment was achieved using gas chromatography (Table 1) of
samples taken from the gas flows using a methaniser and
FID detector system (see Supporting Information for full
details). A standard of CO in CO2 was prepared to calibrate
the system. Duplicate runs were in good agreement, and the
results for [RuCl2(DMSO)4] were similar to those mea-
sured by the IR method. In all cases, the levels of CO in the
gas were between 190 and 440 ppm, with RuCl3 giving the
lowest level and [Ru(NH3)6Cl2] the highest. The isolated
dimer [Ru2(HCO2)2(CO)4], which we speculate may be
the immediate catalyst precursor, produced slightly higher
levels of CO than average, but these were within the
range observed. The close match in the gas composition of

4138 Organometallics, Vol. 28, No. 14, 2009
Table 1. Determination of Level of CO in Gas Flow from Formic Acid Decompositiona
Morris et al.
Ru source
CO rt
CO2 rt
area CO
area CO2
PPM COb
standard ca. 136 ppm CO in CO2
standard ca. 136 ppm CO in CO2
[Ru2(HCO2)2(CO)4]
[Ru2(HCO2)2(CO)4]
[RuCl2(dmso)4]
[RuCl2(dmso)4]
[RuCl3]
[RuCl3]
[Ru(NH3)6Cl2]
0.727
0.742
0.740
0.745
0.729
0.714
0.744
0.736
0.738
0.744
0.738
0.738
1.754
1.787
1.845
1.846
1.827
1.807
1.835
1.835
1.828
1.860
1.825
1.829
0.01281
0.01263
0.03272
0.03274
0.02006
0.01899
0.01958
0.01959
0.04405
0.04224
0.02583
0.02530
99.98719
99.98737
99.96728
99.96726
99.97994
99.98101
99.98042
99.98041
99.95595
99.95776
99.97417
99.97470
128
126c
327
327
200
190
196
196c
440
422
258
253
[Ru(NH3)6Cl2]
[Ru2(PPh3)2(HCO2)2(CO)4]
[Ru2(PPh3)2(HCO2)2(CO)4]
a Using an Agilent 6890N GC fitted with a Porapak Q steel column (6 ft ꢀ 1/8 in. i.d.), methaniser, and FID detector. Oven temperature 50 °C, injector
200 °C, detector 250 °C, gas flow through column 20 mL/min, carrier gas = nitrogen. Gas flow at detector: H2 50 mL/min, air 400 mL/min. Injections of
ca. 0.1 mL. b Duplicates were run within 10 min of each other. c Duplicates were run 24 h apart.
[Ru2(HCO2)2(CO)4] to the other metals suggests that this com-
plex may be formed by each metal in the activation process.
The hydrogen produced in this investigation was fed
directly to a PEM fuel cell and used to power electrical
devices. Although the hydrogen would therefore require
prior purification (for example via gas filtration) for long-
term use in this fuel cell, the advantages of the low catalyst
loading required, the high TOFs, and the long-lived nature of
the catalyst make the process a practical one for large-scale
hydrogen production.
known [RuCl2(PPh3)2(CO)(DMF)], 11 (reported 31P NMR
δ 34.5 ppm).20a DMF-containing mononuclear Ru species
have been reported,20 and DMF is known to be converted to
CO at elevated temperatures in ruthenium complexes.20c
Beller also reported that a similar system could be gener-
ated from RuCl3 and 3 equiv of PPh3 in DMF (80 °C/2 h)
followed by addition of FA/TEA. Again we found that there
was an initial fast hydrogen and CO2 formation after an
induction period. However, analysis of the reaction mixture
prior to the FA/TEA addition revealed the formation of a
predominant product with a characteristic 31P NMR shift at
δ 28.8 ppm. This species appears to be [RuCl3(PPh3)2-
(DMF)] (12), which was independently prepared by the
treatment of RuCl3 with PPh3 in DMF (X-ray details
are given in the Supporting Information). However this
assignment is made with caution, as Ph3PO also appears at
δ 28.8 ppm. Hence, at lower temperatures a less active
monomer complex, possibly 11 or 12, is formed and may
be responsible for FA decomposition at these temperatures.
In a paper that appeared during the preparation of the
manuscript, Laurenczy et al.11b describe the use of a Ru(II)-
based catalyst containing two meta-monosulfonylated tri-
phenylphosphine (TPPMS) ligands and 3-4 water molecules
for the decomposition of a formic acid/sodium formate
mixture. Given that the competing conversion of formic acid
to CO and H2O will generate quantities of water in the
reaction, it is possible that a closely related (i.e., aqua)
complex may have been generated in the Beller system. Such
a complex may also be formed under our phosphine-mod-
ified conditions described above. However, given that the
Laurenczy system operated much more slowly in the absence
of a phosphine, we believe that a distinct species from this is
formed under our phosphine-free conditions.
We also attempted to gather evidence for the in situ forma-
tion of [Ru2(HCO2)2(CO)4] by 13C NMR spectroscopy. Laur-
1
enczy et al. have shown that in situ H NMR of a Ru(II)
complex used in formic acid/sodium formate solution per-
mitted the identification of ruthenium hydride containing
species.11b Unfortunately, we were not able to observe for-
mation of metal hydrides in our catalytic system during the
course of the reaction. We were, however, able to observe
peaks at ca. δ 172 and 204 corresponding to the formate
and carbon monoxide ligands, respectively, upon stirring of
[Ru2(HCO2)2(CO)4] in FA/TEA for 10 min at rt. After
heating to 120 °C these peaks disappear, suggesting that
one or more catalytically active species are formed in the
reaction, for which [Ru2(HCO2)2(CO)4] is a precursor. We
are currently investigating this reaction in order to establish
the structure of the active species.
In order to establish why the Beller system works well in
the presence of triphenylphosphine (in contrast to our
catalyst), we examined the complex [RuCl2(PPh3)3], 10.10
At 120 °C (i.e., the conditions used in our study) this complex
gave a high initial release of gas, but this quickly subsided
(in less that 5 min.). Analysis of the reaction mixture by
31P NMR indicated that a complex system of free PPh3, complex
6, RuCl2(PPh3)3, and other species at lower field had formed.
At a lower temperature of 80 °C, however, and using DMF as
a solvent in which to first form an active catalyst, followed by
reaction with FA/TEA at 40 °C, there was evidence of
initially rapid hydrogen and CO2 formation.10a 31P NMR
analysis of the reaction after 2 h reaction in DMF and prior
to addition of FA/TEA clearly indicated that PPh3 had
dissociated (free PPh3 was observed at δ -6.4 ppm). Chemi-
cal shifts attributed to [RuCl2(PPh3)3] at δ 29.3 and 40.9 ppm
were shifted to 27.4 and 54.3 ppm. Analysis of the reaction
mixture at timed intervals after addition of FA/TEA indi-
cated that a complex reaction mixture had formed where
PPh3 recoordinated to give a complex with a characteristic
31P NMR peak at ca. δ 27.0 ppm. After 120 min a new peak at
δ 35.2 ppm appeared, which could be attributed to the
In conclusion, we have found that hydrogen can be
generated from FA using a number of Ru(II) and Ru(III)
catalysts. Our evidence suggests that the precatalyst appears
to be a bisruthenium complex containing bridging formate
ligands and carbonyl ligands. The structure of the active
catalyst involved in the catalytic cycle is not known. The
decomposition of FA to carbon dioxide and hydrogen is very
rapid and is highly selective, although low levels of CO
(ca. 0.02 mol %) were detected in the outflows. Temperatures
of >100 °C are required to form the dimeric catalysts that
ꢀ
ꢀ
(20) (a)Gomez-Benıtez, V.;Olvera-Mancilla, J.;Hernandez-Ortega, S.;
´
Morales-Morales, D. J. Mol. Struct. 2004, 689, 137–141. (b) Levason, W.;
Quirk, J. J.; Reid, G. Acta Crystallogr. 1997, C53, 1224–1226. (c) Serp, P.;
Hernandez, M.; Richard, B.; Kalck, P. Eur. J. Inorg. Chem. 2001, 2327–2336.

Article
Organometallics, Vol. 28, No. 14, 2009 4139
were studied in this project. At lower temperatures than this,
and in our hands, less active monomeric ruthenium com-
plexes appear to be formed. In view of the close analogy
of alcohol dehydrogenation process with that of FA
dehydrogenation,21 we are extending our studies to the
dehydrogenation of alcohols commonly represented in bio-
mass and waste organic materials.
This will be the same for each 50 mL gas sample in any
particular run. Multiplying this by 20 ꢀ no. liters hydrogen
produced overall gives the overall TON for the run. This
figure will continue to increase as long as the reaction runs.
The turnover frequency was calculated by
TON ꢀ 60
TOF h-1
¼
time=min
Experimental Section
The graphs therefore show “instantaneous” TOFs; in some
cases the graphs (see Supporting Information) are smoothed
by taking the average of the values either side of each data
point, twice (denoted by “averaged twice”). In other cases,
the moving average trend line (3 points) has been applied to
the graph. Graphs of results obtained are given in the
Supporting Information.
Identification of Complex 6 in FA Decomposition Reaction and
Isolation for X-ray Analysis. Ruthenium trichloride hydrate
(100 mg, 0.41 mmol) and triphenylphosphine (54 mg, 0.21
mmol) were added to 5:2 FA/TEA (25 mL), and the reaction
mixture was heated to 100 °C. After 1 h, formic acid (10 mL) was
replenished and the reaction mixture was stirred at 100 °C for a
further hour. The crude reaction mixture was allowed to cool to
room temperature and yielded a yellow crystalline material.
X-ray crystallography revealed structure 6 (see Supporting
Information for full data). 31P NMR of the crude reaction
mixture revealed the formation of a predominant species at
12.6 ppm. The method for formation of a sample for full
characterization is given below.
Formation of 6 from RuCl2DMSO4 under Reaction Condi-
tions. [Ru(DMSO)4Cl2] (40.0 mg, 82.6 μmol) and PPh3 (10.8 mg.
41.3 μmol) were added to TEA/FA (10 mL) and heated at 120 °C
using an oil bath. The two species identified by 31P NMR match
the chemical shifts for [Ru2(O2CH)(CO)4(PPh3)2] (12.6 ppm)
and triphenylphosphine oxide (28.8 ppm). Both sources of
ruthenium (oxidation levels II and III) give the same species
when treated with PPh3 in hot TEA/FA, namely, [Ru2(O2CH)-
(CO)4(PPh3)2].
Evidence of Hydrogen and Carbon Dioxide Formation. In
order to confirm that the gas was a combination of hydrogen
and carbon dioxide, we captured samples in a gas phase cell for
FT-IR spectroscopy (sample spectra are shown in the Support-
ing Information). The spectra showed a strong peak at ca.
2300 cm-1 22
There was evidence of carbon monoxide
.
(2150 cm-1) at low levels.22 The presence of hydrogen was
confirmed by (i) performing a reduction of an NdN double
bond in the dye disperse red 13 (resulting in loss of color) using a
Pd/C catalyst with the gas generated from the reaction flowed
through the stirred reaction and (ii) diverting the gas into a PEM
fuel cell, which was used to power an electric fan.
Procedure for Hydrogen Generation from Formic Acid. An oil
bath was preheated to 131 °C, and [Ru(DMSO)4Cl2] (6.0 mg,
12.4 μmol) was added to a flask to which FA/TEA (5:2; 10 mL)
was added. The flask was connected to a condenser, which in
turn was connected to a gas volume measuring kit containing
two syringes and a three-way tap, which was checked carefully
for leaks. The flask was lowered into the oil bath, and the
stopwatch and syringes were simultaneously set to take mea-
surements. When the internal temperature of the reaction
mixture reached 120 °C (∼3.5 min), the volume of expansion
and any gases given off were noted with the time taken.
Simultaneously, the timer was reset and times taken for each
50 mL of gas generated were recorded. Formic acid was
replenished as necessary, i.e., when activity was seen to signifi-
cantly decay. At this point the volume of formic acid required
was calculated on the basis of the amount of gas generated. For
example, 3 L of gas produced represents 3 L/48 L moles of the
two gases = 62.5 mmol of formic acid consumed. Therefore,
2.88 g (2.36 mL; d =1.22 g/mL) of formic acid was required to
return the reaction mixture back to the original azeotropic ratio.
The turnover number was calculated for every 50 mL generated by
Formation of Precursor to Complex 6. Catena-di-μ-formato-
dicarbonylruthenium was prepared by methods described in the
literature.19 Dodecacarbonyltriruthenium (0.58 g, 2.72 mmol on
Ru) was added to formic acid (25 mL), and the reaction mixture
was heated under reflux for 6 h. Concentration under reduced
pressure gave a pale orange precipitate, which was washed
with ether and dried (0.44 g, 2.16 mmol on Ru, 79%). To a
suspension of thus formed catena-di-μ-formatodicarbonyl-
ruthenium (0.30 g, 0.74 mmol) in THF (30 mL) was added
triphenylphosphine (0.39 g, 1.48 mmol), and the reaction mix-
ture was heated under reflux for 6 h. The reaction mixture was
then cooled to rt and filtered to give a golden yellow solution.
Concentration under reduced pressure and crystallization from
DCM (20 mL) and MeOH (1 mL) three times gave elongated
yellow crystals (0.50 g, 0.54 mmol, 73%). Ten-minute X-ray
crystallography analysis gave a matched data set to 6: mp(dec)
>153.7 °C; νmax/cm-1 (solid) 2022.0, 1979.3, 1945.1, 1920.8,
1581.8, 1480.5, 1434.3, 1354.8; found HRMS (ESI): 930.9755
[MH]þ, C42H33O8P2Ru2 requires 930.9732 (3.42 ppm error);
m/z (ESI) [M]þ 929.99; δH (300 MHz; CDCl3; Me4Si) 7.36-7.44
(18H, m, ArH), 7.51-7.58 (12H, m, ArH), 8.14 (2H, s, O2CH);
moles gas produced
TON ¼
moles catalyst used
ð0:05 L=48 mol L-1
Þ
TON ¼
moles catalyst used
(21) (a) Fujita, K.-I.; Tanino, N.; Yamaguchi, R. Org. Lett. 2007, 9,
109–111. (b) Adair, G. R. A.; Williams, J. M. J. Tetrahedron Lett. 2005, 46,
8233–8235. (c) Junge, H.; Beller, M. Tetrahedron Lett. 2006, 46, 1031–
1034. (d) Morton, D.; Cole-Hamilton, D. J. Chem. Commun. 1988, 1154–
1156. (e) Ligthart, G. B. W. L.; Meijer, R. H.; Donners, M. P. J.; Meuldijk, J.;
Vekemans, J. A. J. M.; Hulshof, L. A. Tetrahedron Lett. 2003, 35, 1507–
1509. (f ) Dobson, A.; Robinson, S. D. Inorg. Chem. 1977, 16, 137–142. (g)
Morton, D.; Cole-Hamilton, D. J.; Utuk, I. D. J. Chem. Soc., Dalton Trans.
1989, 489–495. (h) Zhang, J.; Gandelman, M.; Shimon, L. J. W.; Rozenberg,
H.; Milstein, D. Organometallics 2004, 23, 4026–4033. (i) van Buitjtenen, J.;
Meuldijk, J.; Vekemans, J. A. J. M.; Hulshof, L. A.; Kooijman, H.; Spek, A. L.
Organometallics 2006, 25, 873–881. ( j) Gunanathan, C.; Milstein, D.
Angew. Chem. Int. Ed. 2008, 47, 8661–8664. (k) Gunanathan, C.;
Ben-David, Y.Milstein, D. Science 2007, 317, 790–792. (l) Hashiguchi, S.;
Fujii, A.; Haack, K.-J.; Matsumura, K.; Ikariya, T.; Noyori, R. Angew. Chem.,
Int. Ed. 1997, 109, 300–303. (m) Arita, S.; Kioke, T.; Kayaki, Y.; Ikariya, T.
Angew. Chem., Int. Ed. 2008, 47, 2447–2449. (n) Royer, A. R.; Rauchfuss, T.
B.; Wilson, S. R. Inorg. Chem. 2008, 47, 395–397. (o) Nordstrøm, L. U.; Vogt,
H.; Madsen, R. J. Am. Chem. Soc. 2008, 130, 17672–17673.
δ
C (75.5 MHz; CDCl3; Me4Si) 128.3 (t, J 4.6), 129.8, 133.1 (t. J
16.4), 133.7 (t, J 5.9), 176.3 (t, J 8.2), 204.6 (t, J 4.0); δP (121.5
MHz; CDCl3; Me4Si) 12.6.
Formation of Complex 7. To a suspension of catena-Di-μ-
formatodicarbonylruthenium (0.50 g, 1.24 mmol) in THF
(60 mL) was added triphenylphosphine (0.24 g, 0.93 mmol),
and the reaction mixture was heated under reflux for 1 h. The
reaction mixture was then cooled to rt and concentrated under
reduced pressure. Isolation by silica gradient flash column
chromatography 1:9-1:0 EtOAc/hexane and crystallization
(22) Krebs, A.; Cholcha, W.; Miller, M.; Eisher, T.; Pielartzik, H.;
Schnockel, H. Tetrahedron Lett. 1984, 44, 5027–5030.

4140 Organometallics, Vol. 28, No. 14, 2009
Morris et al.
from EtOAc (7 mL) and hexane (20 mL) gave light brown
double-ended needles (0.11 g, 82.5 μmol, 9%). X-ray crystal-
lography revealed structure 7. δP (121.5 MHz; CDCl3; Me4Si) 5.8.
Formation of an Authentic Sample of Mono-Ru Complex
[RuCl3(PPh3)2(DMF)], 12. To a solution of triphenylphosphine
(970 mg, 3.70 mmol) in dimethylformamide (60 mL) was added
ruthenium trichloride hydrate (300 mg, 1.23 mmol). The reaction
mixture was stirred at 80 °C for 2 h, concentrated under reduced
pressure, and recrystallized from dimethylformamide (12 mL)
twice. δP (121.5 MHz; CDCl3; Me4Si): 28.8 (NMR spectrum also
shows free triphenylphosphine). Crude X-ray data indicated
atom connectivity matched that of [RuCl3(PPh3)2(DMF)]; how-
ever crystal decomposed during collection and the data were not
of high enough quality to refine.
Acknowledgment. We thank the EPSRC for funding
(postdoctoral grant EP/D031168/1 to D.J.M.). We ac-
knowledge the use of the EPSRC Chemical Database
Service at Daresbury.23 We thank Colin Murrell
and Gerald Chapman of the Department of Biological
Sciences for access to GC facilities for gas measurements.
The Oxford Diffraction Gemini instrument and the GC
facilities used in this research was obtained through the
Science City Project with support from the Advantage
West Midlands (AWM) Advanced Materials Project and
partly funded by the European Regional Development
Fund (ERDF).
Supporting Information Available: Full experimental details
and graphs of experimental results. This material is free of
charge via the Internet at http://pubs.acs.org.
(23) Fletcher, D. A.; McMeeking, R. F.; Parkin, D. J. Chem. Inf.
Comput. Sci. 1996, 36, 746–749.